Experts talk on AI technology

Deep synthesis must be kept in check to prevent its misuse

By Shuai Yinan

Deep synthesis technology is a widely used application in the field of artificial intelligence, particularly in audio and video production, media communication and information services.

One of the most controversial applications of the technology is deepfakes, where synthetic media is used to swap the face or voice of one person for another. While this application demonstrates the value of technology and its commercial potential, it is a double-edged sword, as it will not only benefit the world but also create risks for people's interests, the cyber ecology and social order.

Several cases of AI fraud, in which artificial intelligence tools were used to mimic people's voices and appearance, have sparked concern about the potential of artificial intelligence techniques to aid crime.

In a case recently revealed by the public security bureau of Baotou, Inner Mongolia autonomous region, the victim was tricked into initiating a total transfer of 4.3 million yuan ($612,000) via a company account after having a video chat with someone masquerading as his friend.

Also, some face-swapping photos and videos are posted online to spread disinformation and violate personal privacy. This fake content is used to guide public opinion and trigger a crisis of confidence as it undermines the credibility of official media and government authorities.

Since deep synthesis can be applied to text, voices, images and videos, people will start to doubt everything in an environment where it is difficult to distinguish between true and false. Lawbreakers can easily create fake news for people who are more likely to be imprisoned in information cocoons. Worse, the malicious use of deep synthesis technology to create fake political information may threaten national security.

Many countries and blocs including the United States and the European Union have introduced policies and laws to regulate the application of the deep synthesis technology.

China has also responded to legal problems raised by the technology. In November, the Cyberspace Administration of China, the Ministry of Industry and Information Technology and the Ministry of Public Security jointly issued the provisions for the administration of the use of deep synthesis in the field of internet information services, imposing clear constraints on the generation, replacement and manipulation of human faces, as well as on the synthesizing and mimicking of human voices.

The development of deep synthesis technology is at a critical period so that it needs to be supervised and managed to avoid it causing any harm to society.

A standard supervision mechanism should be established to evaluate digital content due to its easy accessibility, mobility and wide influence.

The rapid development of deep synthesis technology requires advanced identifying and decoding technologies to avoid risks caused by the misuse of technology. Efforts should be made to research technologies such as digital watermarking, watermarking algorithms and traceability for content distribution in a bid to identify digital works.

The authorities should also promote supervision through proactive screening. A supervision platform needs to be established to monitor online content and an integrated human-computer supervisory architecture should be introduced to evaluate the veracity of online content. This kind of bi-channel control can reduce legal risks caused by disinformation.

The improvement of ethical reviews is another way to boost the healthy development of deep synthesis technology. Researchers and service providers should demonstrate the value, morality and safety of their deep synthesis technology.

As such, the evaluation criteria for the technology should be strengthened to support risk assessment. Expert reviews should be introduced to check certain algorithmic models even though it may affect the efficiency of technology commercialization.

In the long run, expert reviews can benefit the development of enterprises as the reviews can help the enterprises avoid ethical and legal risks before a product is put on the market. It is also necessary to enhance ethical awareness so that deep synthesis technology is not misused.

Laws, regulations and standards related to the technology need to be formulated and improved, making clear the scope of its use and the attendant responsibilities. A classification system for the evaluation of the technology also needs to be established to protect national cybersecurity, and legal rights and interests of entities.

Further, there is a need to make clear the definition of improper acts and their legal consequences, and establish a system involving civil law, administrative law and criminal law to cover the research and development and use of deep synthesis technology. A responsible reporting system can be established to manage platforms, guide the development of the technology and encourage market entities to take the initiative to regulate the technology.

The author is a lecturer of politics and law at the Party School of the CPC Central Committee.

The views don't necessarily reflect those of China Daily.